Galapagos-2/Galfitm/GAMA -- multi-wavelength measurement of galaxy structure: separating the properties of spheroid and disk components in modern surveys

Résumé : We present the capabilities of Galapagos--2 and Galfitm in the context of fitting 2-component profiles to galaxies, on the way to providing complete multi-band, multi-component fitting of large samples of galaxies in future surveys. We release both the code and the fit results to 234,239 objects from the DR3 of the Gama survey, a sample significantly deeper than previous works. We use stringent tests on both simulated and real data, as well as comparison to public catalogues to evaluate the advantages of using multi-band over single-band data. We show that multi-band fitting using Galfitm provides significant advantages when trying to decompose galaxies into their individual constituents, as more data are being used, by effectively being able to use the colour information buried in the individual exposures to its advantage. Using simulated data, we find that multi-band fitting significantly reduces the deviations from real parameter values, allows component sizes and S\'ersic indices to be recovered more accurately, and, by design, constrains the band-to-band variations of these parameters to more physical values. On both simulated and real data, we confirm that the SEDs of the 2 main components can be recovered to fainter magnitudes compared to using single-band fitting, which tends to recover disks and bulges to - on average - have identical SEDs when the galaxies become too faint, instead of the different SEDs they truly have. By comparing our results to those provided by other fitting codes, we confirm that they agree in general, but measurement errors can be significantly reduced by using the multi-band tools developed by the MegaMorph project. We conclude that the multi-band fitting employed by Galapagos-2 and Galfitm significantly improves the accuracy of structural galaxy parameters and enables much larger samples to be be used in a scientific analysis.